tidy up dither stage

Using the float iota was just an expedient to write the stage...
this CL adds the U32 iota that dither really wants.

Change-Id: I7990b10afd0c5277186b6b8e730245d291bcef0c
Reviewed-on: https://skia-review.googlesource.com/17441
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>

src/jumper/SkJumper.cpp

@@ -22,7 +22,8 @@
 // It's fine to rearrange and add new ones if you update SkJumper_constants.
 using K = const SkJumper_constants;
 static K kConstants = {
-    {0.0f, 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f, 7.0f},
+    {0,1,2,3,4,5,6,7},
+    {0,1,2,3,4,5,6,7},
 };
 
 // We can't express the real types of most stage functions portably, so we use a stand-in.

src/jumper/SkJumper.h

@@ -51,7 +51,8 @@
 static const int SkJumper_kMaxStride = 8;
 
 struct SkJumper_constants {
-    float iota[SkJumper_kMaxStride];   //  0,1,2,3,4,...
+    float    iota_F  [SkJumper_kMaxStride];   //  0,1,2,3,4,...
+    uint32_t iota_U32[SkJumper_kMaxStride];   //  0,1,2,3,4,...
 };
 
 struct SkJumper_GatherCtx {

src/jumper/SkJumper_stages.cpp

@@ -286,7 +286,7 @@ STAGE(seed_shader) {
     // It's important for speed to explicitly cast(x) and cast(y),
     // which has the effect of splatting them to vectors before converting to floats.
     // On Intel this breaks a data dependency on previous loop iterations' registers.
-    r = cast(x) + 0.5f + unaligned_load<F>(k->iota);
+    r = cast(x) + 0.5f + unaligned_load<F>(k->iota_F);
     g = cast(y) + 0.5f;
     b = 1.0f;
     a = 0;
@@ -297,7 +297,7 @@ STAGE(dither) {
     auto c = (const SkJumper_DitherCtx*)ctx;
 
     // Get [(x,y), (x+1,y), (x+2,y), ...] loaded up in integer vectors.
-    U32 X = trunc_((int)x + unaligned_load<F>(k->iota)),  // Going through float is kind of lazy..
+    U32 X = x + unaligned_load<U32>(k->iota_U32),
         Y = (uint32_t)*c->y;
 
     // We're doing 8x8 ordered dithering, see https://en.wikipedia.org/wiki/Ordered_dithering.